Learning-Based Cell Injection Control for Precise Drop-on-Demand Cell Printing.

Area of Science:

• Bioprinting and Tissue Engineering
• Computational Fluid Dynamics
• Machine Learning Applications

Background:

• Drop-on-demand (DOD) printing is crucial for bioprinting in tissue engineering due to its minimal cell damage and cost-effectiveness.
• Satellite droplets in DOD printing hinder positional accuracy and cell placement.
• Current methods for controlling cell injection in DOD printing rely on inefficient trial-and-error processes.

Purpose of the Study:

• To develop a novel machine learning approach, Learning-based Cell Injection Control (LCIC), for precise DOD bioprinting.
• To automatically eliminate satellite droplets and optimize printing parameters.
• To enhance printing accuracy and efficiency for manufacturing complex artificial tissues.

Main Methods:

• Utilized a computational fluid dynamics (CFD) simulation model of a piezoelectric DOD print-head, incorporating the inverse piezoelectric effect.
• Employed a multilayer perceptron (MLP) network, trained on simulation data generated by the CFD model.
• Leveraged artificial neural network algorithms to optimize DOD printing parameters automatically.

Main Results:

• Achieved a test accuracy of 90% for the LCIC method.
• Significantly reduced satellite droplets in piezoelectric DOD printing through experimental validation.
• Demonstrated substantial improvements in printing efficiency and precision.

Conclusions:

• The LCIC method offers an effective, automated solution for controlling cell injection in DOD bioprinting.
• This approach overcomes the limitations of trial-and-error methods, improving manufacturing precision for artificial tissues.
• The LCIC framework has potential for further optimization of DOD print-head design and cell behavior control.